Human—wildlife interactions are often associated with a myriad of stakeholder groups, intense political scrutiny, and limited biological data, creating complex decision-making situations for wildlife management agencies. Limited research exists on the development and testing of tools (e.g., models to predict the spatial distribution of interactions) to reduce human—black bear (Ursus americanus) interactions (HBI). Available models predicting spatial distribution of HBI are usually developed at scales too large to predict across urban areas, are rarely tested against independent data sets, and usually do not incorporate both landscape and anthropogenic variables. Our objective was to develop a predictive modeling tool that could identify areas of high conflict potential across urban landscapes. We compared locations of HBI in Missoula, MT, recorded by Montana Fish, Wildlife & Parks from 2003 to 2008, to random locations using logistic regression. The final model discriminated the relative spatial probability of HBI within Missoula well, and a second study area moderately. The probability of HBI in Missoula increased when residents lived close to forested patches and major rivers and streams and in intermediate housing densities (approx. 6.59 houses/ ha). Our results provide a wildlife management tool and a repeatable statistical framework that predicts spatial distribution of HBI using only a small set of variables.